AI INSIGHTS DAILY
Powered by Vedya Generative AI
Microsoft Lets Free Copilot AI Users Use Voice Conversation and Complex Query Tools
25 February 2025
Microsoft’s Copilot now provides free users with unlimited access to Voice and Think Deeper features. Voice permits verbal interaction, while Think Deeper enables more intricate queries. These features, which include a male British accent option, are accessible to free users but may be restricted during peak hours.
OpenAI expands Deep Research access to Plus users, heating up AI agent wars with DeepSeek and Claude
25 February 2025
OpenAI’s Deep Research tool, using the upcoming o3 model, is now accessible to ChatGPT Plus, Team, Education, and Enterprise users. This AI agent scours online sources, analyzes text, images, and PDFs, and generates comprehensive reports, challenging the closed, subscription-based business model of Western AI development.
Tesla AI Domination Starts in 2025
24 February 2025
Tesla and XAI are building data centers at a much faster rate than competitors like Microsoft, Google, and Meta, which is giving them a competitive advantage in AI. This speed and scale are also seen in SpaceX’s reusable booster Falcon 9 and Starship.
Thinking Machines Lab is OpenAI ex-CTO Murati's new AI startup
18 February 2025
Mira Murati, OpenAI’s former CTO, has unveiled her new AI startup, Thinking Machines Lab. The startup focuses on AI safety, transparency, and infrastructure quality. It aims to build “multimodal systems that work with people collaboratively” and prioritizes infrastructure quality and preventing the misuse of models.
Meta Announces Dates for Connect 2025
18 February 2025
Meta Connect, Meta’s annual developer conference, is scheduled for September 17-18, 2025. The event will highlight advancements in AI, VR, and the metaverse, with CEO Mark Zuckerberg sharing Meta’s future vision. Attendees can anticipate new AI updates, VR/AR technology showcases, and networking opportunities.
Google Search's AI Mode to be available just a tap away on Pixels
17 February 2025
Google is reportedly developing an AI Mode for Google Search, which may be accessible on Pixel phones via an updated home screen widget or the Google app. The AI Mode aims to provide quicker and more direct answers, potentially saving users from browsing through search results. The specific release date and availability on other Android devices remain uncertain.
An AI Retrospective: From Hype To Reality
14 February 2025
Michael Wegmüller, Co-Founder of Artifact SA, discusses the advancement of generative AI (GenAI) in 2024, including improved reasoning capabilities of LLMs, specialized AI agents, and AI-driven video generation tools.
Robotaxis At A Crossroads 20 Years After DARPA
14 February 2025
Xingjian “XJ” Zhang, Head of Growth at Apex.AI, discusses the present state of the autonomous vehicle (AV) industry.
New Research Suggests AI Assistants Could Be Impact- ing Your Cognitive Skills
14 February 2025
A Microsoft-Carnegie Mellon University study reveals that overreliance on AI tools like Copilot and ChatGPT may lead to a decline in critical thinking abilities due to the reduction of opportunities for judgment practice. The research aligns with users’ concerns about cognitive engagement, emphasizing the importance of balancing AI use with maintaining cognitive skills.
Google opens door to AI-powered weapons, surveillance Google
30 January 2025
Google has updated its AI principles, potentially allowing the company to engage in AI-powered weapons and surveillance systems, a shift from its earlier prohibition. The updated principles emphasize bold innovation, responsible development, and collaborative progress.
4 Warnings About deepSeek you Need to Know Before Using It
30 January 2025
Security experts and AI professionals have raised concerns about DeepSeek’s data retention, privacy policy, censorship, and energy consumption. DeepSeek retains user data indefinitely, tracks keystrokes, censors outputs, and may not be more energy-efficient than existing models. Organizations are advised to consider these risks when deploying DeepSeek at scale.
The future of generative AI: 10 trends to follow in 2025
30 January 2025
Here are 10 GenAI trends for 2025:
1. Executives seek ROIs from GenAI investments. 2. GenAI initiatives will increasingly add a bottom-up approach. 3. GenAI will make everyone a programmer. 4. Organizations and executives will build trust in GenAI systems.
Ai spy! India keeps vigil as privacy fears run deep
30 January 2025
Indian officials are monitoring the DeepSeek AI model, developed by Chinese firm Alibaba, due to concerns about data safety and sovereignty. The model’s privacy policy allows data storage in China, which has sparked concerns among officials. India had previously banned Chinese apps, including TikTok and WeChat, due to espionage concerns.
Tesla Teases Unsupervised Full Self-Driving Cars From Factory To Loading Dock
30 January 2025
Tesla has shared a video showcasing its electric vehicles driving autonomously from their factory to designated loading docks, without human intervention. The company is working towards large-scale unsupervised Full Self-Driving (FSD), although there’s still a significant gap between navigating controlled circuits and public roads.
The stable version of Gemini 2.0 Flash is rolling out now in the Gemini app
30January 2025
Google is releasing Gemini 2.0 Flash, a stable version of its AI model, for free users in the Gemini app, replacing the experimental version. Google One AI Premium subscribers still get exclusive perks. The model offers faster responses and better performance. Gemini 1.5 Flash and Pro will be available for a few more weeks before being replaced.
AI 'godfather' Yoshua Bengio says AI agents could be the 'most dangerous path'
24 January 2025
Yoshua Bengio, one of the “AI godfathers,” has warned against the dangers of AI agents, which could lead to catastrophic scenarios. Bengio is researching non-agentic systems to keep agents in check and has called for national regulation to prevent AI companies from building agentic models without proving safe
AI could surpass “almost all humans" by 2027, Anthropic chief says
23 January 2025
Anthropic CEO Dario Amodei said that AI could surpass “almost all humans” in the next couple of years. Amodei raised concerns about the role of humans once AI and robotics can replace workers in all possible fields. He thinks that when the time comes, we will have to completely rethink the way people view work, themselves, and others.
How Cisco leverages AI for impact | FY24 Purpose Report
23 January 2025
Cisco integrates AI into its strategy, products, and services. It uses AI to support employees, advance skills, and enhance security response. Cisco launched BridgeIT, a Cisco Enterprise Chat AI, and piloted “Teaming with AI” to improve productivity. It offers AI learning content and led the launch of the AI-Enabled ICT Workforce Consortium.
Hugging Face showcases How Test-Time Compute Scaling Can help SLM's Outperform Larger AI
24 December 2024
Hugging Face researchers found that allocating more time for AI models to solve problems and self-correct during testing can improve performance without increasing training time or data.
Meta is in a spot of bother as ChatGPT comes to WhatsApp
24 December 2024
Meta’s WhatsApp is feeling the heat from the integration of ChatGPT into the messaging platform. This development could lead to a significant shift in the chatbot market, with potential implications for Meta’s dominance.
OpenAI's o3 suggests AI models are scaling in new ways - but so are the costs | TechCrunch
23 December 2024
OpenAI’s o3 model, which uses more compute during ChatGPT’s inference phase, significantly outperforms other AI models on benchmarks, indicating progress in AI scaling laws.